START Find a list of possible correspondences between two obj Sort out 3-tuples(selection process) Search for the optimal 3-tuple END Figure 21. 1: A diagram of the KH method is selected to satisfy the following Euclidean distance constraints simultaneously m1-m2|-|n1-n2||<6 - n1< 21.1 where Sselect is a user-defined tolerance Step 14 The correspondence information between each point m; on ri and n: on r2 is established from which a list of translation vectors and a rotation matrices can be obtained. We choose a translation vector and a rotation matrix which produces a minimum value of equation(21.4 with s= 1, see Ko et al.[8 21.8 Matching problems: NGWs Since global matching is considered, scaling effects in the matching problem of this type can be easily resolved by using integral properties such as surface areas or volumes. After scaling transformation, the center of mass and the principal moments are used to find the rigid bodyFind a list of possible correspondences between two objects. Sort out 3−tuples (selection process). Search for the optimal 3−tuple. 10 12 14 START END Figure 21.1: A diagram of the KH method is selected to satisfy the following Euclidean distance constraints simultaneously ||m1 − m2| − |n1 − n2|| < δselect, ||m2 − m3| − |n2 − n3|| < δselect, ||m3 − m1| − |n3 − n1|| < δselect, (21.12) where δselect is a user-defined tolerance. Step 14 The correspondence information between each point mi on r1 and ni on r2 is established, from which a list of translation vectors and a rotation matrices can be obtained. We choose a translation vector and a rotation matrix which produces a minimum value of equation (21.4) with s = 1, see Ko et al. [8]. 21.8 Matching problems : NGWS Since global matching is considered, scaling effects in the matching problem of this type can be easily resolved by using integral properties such as surface areas or volumes. After scaling transformation, the center of mass and the principal moments are used to find the rigid body transformation. 8